English

Dimension-Free Noninteractive Simulation from Gaussian Sources

Probability 2022-02-21 v1 Computational Complexity Information Theory math.IT

Abstract

Let XX and YY be two real-valued random variables. Let (X1,Y1),(X2,Y2),(X_{1},Y_{1}),(X_{2},Y_{2}),\ldots be independent identically distributed copies of (X,Y)(X,Y). Suppose there are two players A and B. Player A has access to X1,X2,X_{1},X_{2},\ldots and player B has access to Y1,Y2,Y_{1},Y_{2},\ldots. Without communication, what joint probability distributions can players A and B jointly simulate? That is, if k,mk,m are fixed positive integers, what probability distributions on {1,,m}2\{1,\ldots,m\}^{2} are equal to the distribution of (f(X1,,Xk),g(Y1,,Yk))(f(X_{1},\ldots,X_{k}),\,g(Y_{1},\ldots,Y_{k})) for some f,g ⁣:Rk{1,,m}f,g\colon\mathbb{R}^{k}\to\{1,\ldots,m\}? When XX and YY are standard Gaussians with fixed correlation ρ(1,1)\rho\in(-1,1), we show that the set of probability distributions that can be noninteractively simulated from kk Gaussian samples is the same for any km2k\geq m^{2}. Previously, it was not even known if this number of samples m2m^{2} would be finite or not, except when m2m\leq 2. Consequently, a straightforward brute-force search deciding whether or not a probability distribution on {1,,m}2\{1,\ldots,m\}^{2} is within distance 0<ϵ<ρ0<\epsilon<|\rho| of being noninteractively simulated from kk correlated Gaussian samples has run time bounded by (5/ϵ)m(log(ϵ/2)/logρ)m2(5/\epsilon)^{m(\log(\epsilon/2) / \log|\rho|)^{m^{2}}}, improving a bound of Ghazi, Kamath and Raghavendra. A nonlinear central limit theorem (i.e. invariance principle) of Mossel then generalizes this result to decide whether or not a probability distribution on {1,,m}2\{1,\ldots,m\}^{2} is within distance 0<ϵ<ρ0<\epsilon<|\rho| of being noninteractively simulated from kk samples of a given finite discrete distribution (X,Y)(X,Y) in run time that does not depend on kk, with constants that again improve a bound of Ghazi, Kamath and Raghavendra.

Keywords

Cite

@article{arxiv.2202.09309,
  title  = {Dimension-Free Noninteractive Simulation from Gaussian Sources},
  author = {Steven Heilman and Alex Tarter},
  journal= {arXiv preprint arXiv:2202.09309},
  year   = {2022}
}

Comments

33 pages

R2 v1 2026-06-24T09:44:51.696Z